Add argument acc.davies for pedgene. Some kernel tests gave NA p-values with default daves accuracy of 1e-6. Performs well with default of 1e-5, but give option to user in case NAs occur.
return n-variants for the gene tests, now a column in the returned object from pedgene
Allow the method to complete if any gene has no markers even after all are removed for having zero variance (same genotype)
When a gene has only one marker, the kernel test reduces to burden test. Return same test stat and p-value for both. Thanks to Seth Ament.
Fix bug where failed to subset genotype matrix when either NA in trait or all NA genotypes for a given person, after matching geno and ped subject by ped-person ids. Thanks again to Seth Ament.
Change maintainer to Jason, rather than Dan.
Add n.noninform to pedgene object, the number of non-informative markers that were removed
Require ped and person ID columns in genotype data. To match to more-complete subjects in pedigree data by ped and person ID.
Added argument checkpeds, allows user to turn off the step that performs pedigree checks.
Kinship function (from kinship2) is now called without making a pedigree object first. This step allows disconnected pedigrees, and identifiers can be either numberic or character.
Perform Davies' p-value for Kernel statistic.
Added package vignette, a simple manual
wrapper called pedgene to call pedgene.stats on one gene at a time while re-using kinship matrix for all pedigrees
require ped and geno data.frames to have same number of rows
add test script: test.pedgene.R
if map given (with chrom and gene columns), require map to have as many rows as geno has columns
define S3 pedgene object, with simple print and plot methods
Original release, mostly code from Dan's scripts
pedgene.init pre-calculates some stats, passed to pedgene.stats for one single gene